[Ffmpeg-devel-irc] ffmpeg.log.20141203

[burek]

[04:11] <fatbastard79> I'm having problems with trimming and then either re-encoding or concatenating a file
[04:12] <fatbastard79> It seems that the trim function is losing the codec metadata of the video
[04:16] <fatbastard79> pastebin is here: http://pastebin.com/mriNsnTx
[09:00] <Pkunk> What do i set in AVCodecContext to get the libx264 encoder to spit out keyframes ?
[09:01] <Pkunk> I've set - ctx->gop_size = 25;  ctx->keyint_min = 15;
[09:01] <Pkunk> but still can't find any keyframes in the output
[09:02] <pzich> you want to export keyframes as images? or what?
[09:03] <Pkunk> I'm transcoding an input stream to h.264 + mpeg-ts
[09:03] <Pkunk> but the output has no keyframes so vlc/ffmpeg is unable to detect the video picture size of the output stream
[09:06] <Pkunk> Stream #0:0[0x100]: Video: h264 ([27][0][0][0] / 0x001B), 90k tbr, 90k tbn, 180k tbc
[09:07] <Pkunk> I've set tbc to 1/25 & tbr to 1/25 .. Somehow the output stream mungles it to invalid values
[09:16] <Mavrik> Pkunk, you need to set frame type for each frame to AV_TYPE_NONE for x264 to automatically determine keyframes
[09:17] <Mavrik> also, x264 will ALWAYS make first frame the keyframe
[09:17] <Mavrik> are you sure your issue isn't somewher eelse?
[09:19] <Pkunk> Mavrik: Thanks for the tip , the input is a series of AVFrames constructed from opencv CV::Mat images I get after running an input video stream through opencv
[09:30] <Pkunk> Mavrik: I have ffmpeg 1.26 & in that AVFrame->pict_type has no AV_PICTURE_TYPE_NONE option
[09:31] <Pkunk> Is there a workaround I can use for ffmpeg 1.26 so that the libx264 encoder generates keyframes automatically ?
[09:44] <Mavrik> Pkunk, just look at doc and see what the actual option is
[09:44] <Mavrik> Pkunk, or look at libx264.c source to see what it expects.
[10:14] <raray> I have a project that already uses libavcodec for reading audio.
[10:14] <raray> I want to encode png files. (Actual images, not movies.) Can libavcodec do it? shouldi use a dedicated library for that?
[10:18] <Pkunk> raray: You'll need a library to decode the .png files to RGB or YUV frame data , and then craft AVFrames which you can pass into libavcodec for encoding
[10:20] <Guest59294> hi
[10:20] <Guest59294> how are you??
[10:22] <raray> Pkunk: ok, will try
[10:48] <k_sze> Is there a good container format for raw YUV video?
[10:48] <Mavrik> raray, yes, you can use libavcodec to do that
[10:48] <Mavrik> you just have to setup the full chain for "png" encoder... maybe using libpng is easier
[10:49] <k_sze> I mean a container format that at least includes the resolution and framerate information so I don't have to pass that information as command line arguments for playback.
[11:45] <Pkunk> How do I lock the bitrate for libx264 ? Encoding some HD content it seems to keep increasing the bitrate with every frame
[11:45] <Pkunk> I'm setting ctx->bit_rate=2000; ctx->bit_rate_tolerance=0;
[11:50] <Pkunk> i mean ctx->bit_rate=2000*1024;
[11:51] <Pkunk> 2000kbps should be enough for 1280x720p network stream .. But it uses way too much bandwidth after a few seconds of encoding
[11:52] <Pkunk> Should i set ctx->coder_type to something ?
[12:08] <Mavrik> Pkunk, you need to set "bit_rate", "max_rate" and "buf_size" to get stable output
[12:08] <Mavrik> don't set tolerance to 0 because there's no such thing as CBR video
[12:08] <Mavrik> you'll just trash all the settings
[12:09] <Mavrik> do tsomething like bit_rate=2000, max_rate=2200, buf_size=1100
[12:09] <Mavrik> Pkunk, alsop
[12:09] <Mavrik> Pkunk, sorry, the units are in BITS/S
[12:09] <Mavrik> so bit_rate = 2000000 to set 2Mbit
[12:09] <Mavrik> if you set it to 2000 it'll just ignore you because you can't stuff video in 2kbit/s :P
[12:12] <Pkunk> Thanks , I missed the max_rate / rc_max_rate members
[12:12] <sacarasc> 2M works as 2Mbps, too.
[12:13] <Pkunk> Yeah I set to 2000 to test if it ignores what I set anyway
[12:29] <waNNaBe> hey, can anybody tell me how to reply in the mailinglist without being s
[12:30] <waNNaBe> .. a subscriber
[12:30] <Mavrik> sacarasc, not in code :P
[12:31] <sacarasc> Oh, sorry.
[14:05] <Pkunk> I'm using ctx->rc_buffer_size & ctx->rc_max_bit_rate to try and keep libx264 output bitrate within 2500kbps
[14:09] <Pkunk> The bitrate stays within 3000kbps now , but the output stream has lots of encode artifacts, like large blocks of pixelation
[14:09] <Pkunk> The source is 1280x720p at 25fps .. It should encode to 2500kbps just fine
[14:12] <Pkunk> Is there some code I'm missing on controlling the output bitrate using libx264 ?
[14:33] <Keestu> hi, i have an generic question. why always it is 30 fps or 60 fps or 120 fps?. why can't it be in between values like 40 or 50 ?
[15:01] <amaurymedeiros> hi! I'd like to know if there are any ffmpeg official ports to android. i'm facing some troubles while testing some ports i found on the web, so i appreciate if anyone can link me to an official or maintained port :)
[15:47] <Keshl> amaurymedeiros: I've looked for the same thing, with no result. I imagine you're trying to convert videos to something your tablet plays, /on/ your tablet. If so, hug Bsplayer instead. oÉo.
[15:47] <Keshl> If'ya do find a maintained and current port, though, lemme know. -É-.
[15:50] <amaurymedeiros> Keshl, actually I want to convert JPG to several mp4 files (each one with a given frame rate) and then merge all videos into one. I'm able to do this using ffmpeg on my pc, but i want to do that in an android app
[15:50] <amaurymedeiros> but i'll let you know if i find something :)
[15:51] <Keshl> Shiny. -É-.
[15:56] <Popara> Hello, it is possible using hls to make the first segment 2 seconds and all the others 10 for example?
[16:02] <anshul_mahe> Popara: you need to change code for that
[16:02] <anshul_mahe> its generally not possible
[16:07] <Nopik_> hi, I have some (probably very rare) use case. Namely I have some icecast audio stream, available to me via http, which I'm downloading and putting to ffserver for further streaming. That part works well. Now, I occasionally, based on some external trigger (I can make it of any form), I want to cut out the current content of audio stream and replace it with
[16:07] <Nopik_> some other content (e.g. another live stream) - everything online, without stopping ffmpeg and knowing timestamps beforehand
[16:09] <Nopik> any idea how to do that in easiest way? I can write some filter, if necessary, but I'd prefer to not go to such low level
[16:12] <Nopik> to describe what I'm trying to do with other words: lets suppose that I have some realtime streaming radio stream, which appears to have occasional commercials. Lets also suppose that I have some system which will recognize those commercials, and can give trigger about start/end of commercial (e.g. run a script, send some network packet, anything). And I
[16:12] <Nopik> have some other audio source, which I want to play instead of commercials, so ffmpeg would need to temporarily switch sources from original to secondary (and then back), based on external triggers - in such way that the output would be continuous
[16:12] <Nopik> any idea how to approach that?
[16:12] <Nopik> ffmpeg might as well just not be able to do that, which is the case, I suppose - but I'd love to know if it is otherwise
[17:09] <anshul_mahe> I am having some different error given by ffmpeg library
[17:09] <anshul_mahe> http://pastebin.com/vuZKWJKA
[17:09] <anshul_mahe> any one know what could be the solution
[17:10] <klaxa|work> what command did you use?
[17:10] <anshul_mahe> it looks like x264 is complaing from inside that correct my setting
[17:10] <anshul_mahe> I am using library
[17:10] <klaxa|work> ah, right
[17:10] <klaxa|work> can you pastebin your code then?
[17:14] <anshul_mahe> It large code, may paste relevent code where error is coming
[17:16] <anshul_mahe> error is coming from avcodec_open2, i pasted in pastebin but its page is moving round nd round
[17:19] <anshul_mahe> ohh done, http://pastebin.com/xrnrjjVe
[17:22] <anshul_mahe> that error is coming from this part of code, i have not set anything related to that
[17:22] <anshul_mahe> http://pastebin.com/yFibZPkw
[17:23] <anshul_mahe> its encoder.c in x264 code base
[17:25] <somebody_useless> I got another problem to report ;), when audio streams change from stereo to Dolby and vice versa ffmpeg stops processing the sound. Volume and silence detect plugins both show appropriate levels in dB. Seems like a processing issue.
[17:35] <anshul_mahe> I did solve my problem by adding these parameter http://pastebin.com/tZxt3C4v
[17:45] <MadTBone> is anyone aware of a web based video player (flash, html5, silverlight.... really anything) that can play yuv422 color subsampling?
[17:51] <MadTBone> also, the same question for players supporting multiple video streams in the container, displayed simultaneously while maintaining time alignment?
[18:32] <mohsen-rashidi> hello there. i have a movie which is divided into several VOB files. how can i unite them and make a mp4 or whatever file using ffmpeg?
[19:03] <ribasushi> hi
[19:04] <ribasushi> any time I use image2pipe I get the following warning: [image2pipe @ 0x8bf260] Stream #0: not enough frames to estimate rate; consider increasing probesize
[19:05] <ribasushi> even though I did specify the framerate explicitly:
[19:05] <ribasushi> ... -r $target_framerate -f image2pipe -c:v ppm -i - ...
[19:05] <ribasushi> is there a way to silence this warning?
[20:20] <quellen> i would like convert a single .jpo to an .avi of 40 minutes
[20:20] <quellen> * .jpg
[20:22] <quellen> itried "ffmpeg -loop 1 -i image.jpg -t 2400 video.avi" but the output file is too big
[20:31] <quellen> :-(
[20:48] <Keshl> I'm trying to sync video with audii in ffmpeg. I have a clapper, and I know what frame the sound need to start at. The sound is in a seporate file. I've tried approaching this by decompiling the video into indivudal frames with the intent to recompile it, but simply removing the leading frames doesn't work. I've tried adding "-ss" before the input field, but that only skips a few frames, not the appropriate number of seconds. Adding -ss
[20:48] <Keshl> after the input fields skips the audio ahead too, thus essentially copping things while keeping the feeds unsynced. How should I approach this? x.x
[20:48] <Keshl> ... Man, I can't type today. x.x
[20:56] <Keshl> In other news, as a kludge, I just took the audio file and added silence before it starts. Still. <É<
[21:01] <talinn> hi, does anyone have any experience with segmenter and hls streams?
[21:07] <quellen> this chat is useless
[21:19] <sakamotoryuichi> hi, does mp4 container requires audio specific config in extradata for aac stream ?
[22:37] <stephenmac7> Question: is there any managed ffmpeg hosting service that anyone knows of?
[22:38] <luc4> Hello! I have a question: I transcoded a h264 high profile video with ffmpeg using these params: -vcodec h264 -crf 22 -preset veryslow -acodec ac3. What I see is that the video went from 1.8G to 949M. The resulting video still is high profile. How can we explain this difference? Some of the params of the new h264 (like cabac or so) that is better implemented in libx264?
[00:00] --- Thu Dec  4 2014